Upgrade_settings error on ridiculously old pipelines

[braymp]

(Via the forum)

The attached pipelines are pretty old (CP 1.0, but pre-9717) and fail upon loading with the following msg:

Traceback (most recent call last):
  File "cellprofiler\pipeline.pyc", line 405, in create_from_handles
  File "cellprofiler\pipeline.pyc", line 435, in instantiate_module
  File "cellprofiler\modules\__init__.pyc", line 387, in instantiate_module
ValueError: Could not find the \ud7c0\udc00\u97c0\udc00\u17c0\udc00\ud7c0\udc00\
u17c0\udc00\u17c0\udc00\u17c0\udc00\u97c0\udc00\u17c0\udc00\u97c0\udc00 module
  File "cellprofiler\gui\pipelinecontroller.pyc", line 226, in do_load_pipeline
  File "cellprofiler\pipeline.pyc", line 548, in load
  File "cellprofiler\pipeline.pyc", line 416, in create_from_handles

I assume it may not be worth going back this far, but it does seem that there are super-old pipelines still floating out there. Perhaps they could just be pointed to the latest MATLAB version of CP1 to upgrade the pipelines, then over to CP2 to get up to date (which is what I did to get around this).

Attachments: http://cellprofiler.org/issues/uploaded/phi_n7/53BP1_RedPIPE.mat http://cellprofiler.org/issues/uploaded/V192WB/ColocalizationPIPE.mat http://cellprofiler.org/issues/uploaded/iEdoID/Cutoff_53BP1_RedPIPE.mat http://cellprofiler.org/issues/uploaded/mGJPr8/Cutoff_RAD51_GreenPIPE.mat http://cellprofiler.org/issues/uploaded/zGeoPq/Cutoff2_53BP1_RedPIPE.mat http://cellprofiler.org/issues/uploaded/D4scYK/Cutoff2_RAD51_GreenPIPE.mat http://cellprofiler.org/issues/uploaded/EkWXtt/RAD51_GreenPIPE.mat

[LeeKamentsky]

From the error message, it looks like the whole file format is not being parsed correctly. I don't think that it's a question of upgrade_settings being at fault. I think we should at least investigate the problem before the next release. There may be a bug that we can fix that will let us load these.

On Mon, Jun 10, 2013 at 12:21 PM, Mark Bray notifications@github.comwrote:

(Via the forum)

The attached pipelines are pretty old (CP 1.0, but pre-9717) and fail upon loading with the following msg:

Traceback (most recent call last): File "cellprofiler\pipeline.pyc", line 405, in create_from_handles File "cellprofiler\pipeline.pyc", line 435, in instantiate_module File "cellprofiler\modulesinit.pyc", line 387, in instantiate_module ValueError: Could not find the \ud7c0\udc00\u97c0\udc00\u17c0\udc00\ud7c0\udc00\ u17c0\udc00\u17c0\udc00\u17c0\udc00\u97c0\udc00\u17c0\udc00\u97c0\udc00 module File "cellprofiler\gui\pipelinecontroller.pyc", line 226, in do_load_pipeline File "cellprofiler\pipeline.pyc", line 548, in load File "cellprofiler\pipeline.pyc", line 416, in create_from_handles

I assume it may not be worth going back this far, but it does seem that there are super-old pipelines still floating out there. Perhaps they could just be pointed to the latest MATLAB version of CP1 to upgrade the pipelines, then over to CP2 to get up to date (which is what I did to get around this).

Attachments: http://cellprofiler.org/issues/uploaded/phi_n7/53BP1_RedPIPE.mat http://cellprofiler.org/issues/uploaded/V192WB/ColocalizationPIPE.mat http://cellprofiler.org/issues/uploaded/iEdoID/Cutoff_53BP1_RedPIPE.mat http://cellprofiler.org/issues/uploaded/mGJPr8/Cutoff_RAD51_GreenPIPE.mat http://cellprofiler.org/issues/uploaded/zGeoPq/Cutoff2_53BP1_RedPIPE.mat http://cellprofiler.org/issues/uploaded/D4scYK/Cutoff2_RAD51_GreenPIPE.mat http://cellprofiler.org/issues/uploaded/EkWXtt/RAD51_GreenPIPE.mat

— Reply to this email directly or view it on GitHubhttps://github.com/CellProfiler/CellProfiler/issues/662 .

[LeeKamentsky]

I imported 53BP1_RedPIPE.mat in Matlab and saved the workspace like this:

save 53BP1_RedPIPE_resaved.mat -v6

I then loaded the pipeline in CellProfiler. There are a few minor errors in the pipeline update, but the pipeline does load successfully. I think that our Matlab file loader has a bug in it (scipy.io.matlab.mio). I might report the bug to them, but I think the work-around of resaving the .mat files is enough of a resolution to close this.

There are some bugs in the way IdentifyPrimaryObjects.upgrade_settings handles thresholding, but I have a feeling that the pipeline is so old that Otsu is always used and the error is easy enough for the user to correct manually.

53BP1_RedPIPE.mat after resave

[LeeKamentsky]

Bug filed with scipy: https://github.com/scipy/scipy/issues/2573